Fig. 1: Technical and biological validation of targeted NanoSeq.

a, Genome-wide SNV burden estimates, as mutations per base pair, for cord blood granulocytes, sequenced using four different fragmentation and library preparation protocols. Error bars show 95% Poisson CIs. Horizontal lines denote the observed burden (solid) and 95% Poisson CIs (dashed) for cord blood granulocytes sequenced by restriction enzyme NanoSeq⁸. Duplex sequencing and NanoSeq burdens are corrected for missed embryonic mutations, as described in ref. ⁸. Enz, enzymatic fragmentation; sonic, sonication; rep, replicate. b, Trinucleotide mutational spectra of single-cell derived cord blood colonies from a previous study⁵⁶ (top), and cord blood granulocytes sequenced using standard duplex sequencing (middle) and whole-genome NanoSeq (bottom). Duplex sequencing and NanoSeq spectra are corrected by the ratio of genomic to observed trinucleotide frequencies. Cosine similarity 95% CIs are calculated by drawing 1,000 random samples from each observed profile, as described in ref. ⁸. c, Linear regression of genome-wide SNV burdens (estimated using targeted NanoSeq) for whole-blood samples from 371 donors against donor age. Points and their associated error bars represent the point estimates and 95% Poisson bootstrapping CIs of passenger mutation burdens for each sample. Slope and intercept of the fitted model (point estimates and 95% CIs) are indicated. One sample was excluded due to the ratio between upper and lower confidence limits being greater than five. d, Mutation counts for each coding mutation consequence (top) and estimated mutant cell fractions (bottom) for 14 genes under significant positive selection in blood. Mutant cell fractions are shown for individuals aged 65–85 whose blood samples were not selected on the basis of their oral epithelium results.